Computational quantitative syntax
Kristina Gulordava, Paola Merlo · Romance languages and linguistic theory · 2020
Abstract Accounting for the constraints on the possible word orders of a sentence in a language and across the world languages is a core challenge for syntactic theory. In the spirit of computational quantitative syntax, in this paper we present quantitative evidence about Universal 18. We show that corpus data confirms a dispreference for the word order combination where adjectives precede but numerals follow the nouns (Adj-N and N-Num). We then investigate if this dispreference is better explained as a constraint expressed at the level of the dominant orders or at the level of individual structures. Corpus counts support the latter interpretation. Finally, we propose a formal model of how this bias against Adj-N-Num orders can be integrated in the grammar.